AppPerf Latency Bands with color schemes!

I haven’t posted lately about any updates I have been working on in regard to my open source application performance monitoring tool call AppPerf. I have been trying to get some time here and there to work on and improve it and add new features that would be useful. I am happy to say that I finally added a new chart on the overview page that should hopefully make it easier to identify when you are having an issue in regards to the performance of your application. I got the inspiration while reading about query bands on the VidixCortex site. Coincidentally, I have decided to call this new visualization, Latency Bands, as they are measuring the latency across different percentiles of data.

Previously, I was using two separate graphs to represent what I think I am able to show in one graph. I also think its much easier to identify when an issue has risen. First I will show what was previously being displayed on the page:

As you can see, I was showing a Latency and Latency distribution chart. The Latency chart was showing 50, 75, 90, and 95th percentiles. Likewise, the distribution graph was showing, well the distribution of requests across all 100 percentiles. At first glance, its hard to tell what is actually going on here, and with AppPerf, I want to make is extremely easy and quick to say, “Hey, I see a problem.”. Also, while these particular percentiles are helpful, what about the 95th and up percentiles that you are missing. Those are also important. We don’t know if the 100th is 5000ms,50,000ms, or even more. Not that helpful.

Now lets introduce the Latency Bands chart:

Quickly looking at this chart, you can see that something is going on at about 11:39 AM, as indicated by the red spike in there. The way this chart works is that it groups the latencies into 10 buckets based on their latencies. As the latency increases for that bucket, so does the color. The color range goes from blue to red, as more requests are grouped into each bucket. If you want to see the breakdown of the latencies, and how many requests there were, you can mouse over each data point and view the following:

Once you find a spike in your latency, you can then drag over that range to zoom in farther, and create a time scope that shows you what was going on with each of your components at that time. Hopefully this feature will allow you to identify more problems quickly, and determine how to fix them!

Shoot me some comments below if this was helpful!


How to create a Sidekiq service application without Rails

Sometimes you want to create a small Sidekiq service to process requests, without all the extra bloat (and memory usage) from a Rails framework. It turns out that creating such service is not all that hard to do. Create your application using the folders and files below as your starting point.


# Gemfile
source ''
ruby "2.5.0"
git_source(:github) do |repo_name|
  repo_name = "#{repo_name}/#{repo_name}" unless repo_name.include?("/")

gem "sidekiq"


:concurrency: <%= ENV["CONCURRENCY"] || 25 %>
 - default


# config/application.rb

ENV["BUNDLE_GEMFILE"] ||= File.expand_path("../../Gemfile", __FILE__)

require "bundler/setup" if File.exist?(ENV["BUNDLE_GEMFILE"])

ENV["RACK_ENV"] ||= "development"

module Example
 class Application
   def self.root
     @root ||= File.dirname(File.expand_path('..', __FILE__))

   def self.env
     @env ||= ENV["RAILS_ENV"] || ENV["RACK_ENV"] || ENV["ENV"] || "development"

   def self.logger
     @logger ||=

require_relative "initializers/sidekiq"


# config/initializers/sidekiq.rb

Sidekiq.configure_server do |config|
  config.redis = { url: ENV["REDIS_URL"] }


# app/workers/my_worker.rb

class MyWorker
  include Sidekiq::Worker

  def perform(*args)
    # Do work here

Once you have created those files, then you can run the following on the command line:

bundle exec sidekiq -e ${RACK_ENV:-development} -r ./config/application.rb -C ./config/sidekiq.yml

If you want to enqueue jobs from another application using Sidekiq, its as easy as:

Sidekiq::Client.enqueue_to "default", MyWorker, "param_1" => "1", "param_2" => "2"

When you have all this working, then you can start adding things like your database connections and other gems and code to the worker. Let me know if this helps anyone or if you run into any problems!

Installing mysql2 Gem error targeting Mac OS X 10.5 or later

I recently tried to install the mysql2 gem and was presented with the following error. I was on El Capitan 10.11.6 and using ruby 1.9.3-p125.

Gem::Installer::ExtensionBuildError: ERROR: Failed to build gem native extension.
/Users/…/.rvm/rubies/ruby-1.9.3-p125/bin/ruby extconf.rb
checking for rb_thread_blocking_region()… yes
checking for rb_wait_for_single_fd()… yes
checking for mysql.h… yes
checking for errmsg.h… yes
checking for mysqld_error.h… yes
creating Makefile


compiling client.c
couldn’t understand kern.osversion `15.6.0’
compiling mysql2_ext.c
couldn’t understand kern.osversion `15.6.0’
compiling result.c
couldn’t understand kern.osversion `15.6.0’
linking shared-object mysql2/mysql2.bundle
couldn’t understand kern.osversion `15.6.0’
ld: -rpath can only be used when targeting Mac OS X 10.5 or later
collect2: ld returned 1 exit status
make: *** [mysql2.bundle] Error 1

Gem files will remain installed in /Users/…/vendor/bundle/ruby/1.9.1/gems/mysql2-0.3.11 for inspection.
Results logged to /Users/…/vendor/bundle/ruby/1.9.1/gems/mysql2-0.3.11/ext/mysql2/gem_make.out

An error occurred while installing mysql2 (0.3.11), and Bundler cannot continue.
Make sure that `gem install mysql2 -v ‘0.3.11’` succeeds before bundling.

After some searching I found out the solution was to install the gem by setting the Mac Deployment Target as so:

MACOSX_DEPLOYMENT_TARGET=10.9 gem install mysql2

Hope this helps someone else in the future!

Open Source Application Performance App

Its been a while since I have posted anything and I thought I would share with you a small project I have started to work on. Its called app_perf and its a little application that is intended to do application performance monitoring. Right now I am only supporting Ruby via the agent gem, but other languages can easily be added to post metrics to this as well. I encourage anyone interested to clone the project, submit any PR’s and help out making this a lot more full featured. Here is the Github link,

Here are some screen shots:

Let me know what you all think!

How to trim a string and fill remaining space in Ruby

So I needed a way to take a string, and depending on its size, shorten it and fill the remaining space with some arbitrary sequence of characters. Ljust would work, except for the fact that it will fill a string up to a given length, but I only needed to do this when over a certain size. Here is an example:

str = "Somereallylongstring"
if str.length > 10
  puts str[0..7] + "…"
  puts str
# => "Somerea..."

This is the basic idea of what I wanted to do. I decide to make it cleaner and override the String class like so:

class String
  def lfill(len = self.length, fill = ".")
    tmp = self[0..(len – 3)] + "..." if self.length – 3 > len
    return tmp || self

nstr = "Somereallylongstring"
puts str.lfill(10)
#=> "Somerea..."


How to merge strings just like ActiveRecord conditions do

While you can achieve the same functionality using sprintf, this may provide a cleaner approach and one that you are more familiar with. This will allow you to build a string the same way you can use ActiveRecord and the :conditions option.

Basically how this works is by overriding the Array class and adding a method to merge the string and values together into unified string! Enough talk, lets see some code:

class Array
  def merge
    statement, *values = self
    expected = statement.count("?")
    provided = values.size
    unless expected.eql?(provided)
      raise "wrong number of bind variables (#{provided} for #{expected}) in: #{statement}"
    bound = values.dup
    statement.gsub("?") { bound.shift }

As you can see, if you do not provide the right number of values for the statement, it will raise and error. Here is how you would use it:

puts ["Hello ?, how are you", "John"].merge
#=> Hello John, how are you

Likewise, you can use variables to hold values:

message = "Hello ?, how are you"
name = "John"
puts [message, name].merge
#=> Hello John, how are you

This will also work with multiple values:

puts ["Hello ?, ? and ?, how are you", "John", "Joe", "Jim"].merge
#=> Hello John, Joe, and Jim, how are you

The only downside to this currently is that you cannot use a ? in the string you are merging, as it will think its a binding character.

NoMethodError on nil.each?

Have you ever have to iterate over an array from say the params hash? You probably have added some code to make sure that data exists in the hash element first before doing the loop. This is a cool little trick to help condense some code. Normally I would write this:

if params[:accounts] and params[:accounts].size > 0
  params[:accounts].each do |account|
    # Do something with account here

As you can see, kind of ugly. Here is the change:

(params[:accounts] || []).each do |account|
  # Do something with account

AHH!!! Much nicer!

Moving files in Rails using String

Since ruby allow for really easy overriding of objects and classes, I decided to make it easier and cleaner to move my files. Here is how you currently do it:

FileUtils.move("/path/to/file", "/path/to/new/file")

I know this isn’t a huge deal, but I would like to clean it up a little bit. I used an initializer (RAILS_ROOT/config/initializers/core_extensions.rb) to override the String class as such:

class String
  def move(to = nil)
    if to && File.exist?(self)
      FileUtils.move(self, to)

This allows me to do the following to move files:


I though it was pretty cool!

Easy Hash to XML and back conversion in Ruby

Instead of spending hours trying to write something custom to convert your hash into xml or vice versa, you can do it in one line. If you are already in a Rails environment, you are good to go, but if using ruby, include the ActiveSupport gem:

require 'rubygems'
require 'active_support'

Now that you have, we can convert a xml to a hash:

hash = Hash.from_xml(xml_stream)

And a hash to xml is stupid simple:

xml = hash.to_xml

There you have it, simple xml to hash and hash to xml!

Capistrano, CVS, and connection refused issue

If you ever find yourself using CVS as your repository, and are trying to deploy your projects via capistrano, you may run into a connection refused issue. The easiest way I found a solution around this was to modify the capistrano file: “/capistrano-2.1.0/lib/capistrano/recipes/deploy/scm/cvs.rb”. This of course is in your ruby gems folder.I could have added just the change to my project, but I am lazy, and need this for multiple CVS project deploying from this machine. The change I made was on line 145, which I added the CVS_RSH=ssh line:

"export CVS_RSH=ssh && mkdir -p #{ dest } && cd #{ dest }"